{"id":"W4404620644","doi":"10.1039/d4sc04401k","title":"Assessment of fine-tuned large language models for real-world chemistry and material science applications","year":2024,"lang":"en","type":"article","venue":"Chemical Science","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":48,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto; University of Waterloo","funders":"NCCR Catalysis; H2020 European Research Council; National Institute of Diabetes and Digestive and Kidney Diseases; H2020 Marie Skłodowska-Curie Actions; Government of the United Kingdom; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; University of Toronto; Agencia Estatal de Investigación; National Center of Competence in Research Materials’ Revolution: Computational Design and Discovery of Novel Materials; Seventh Framework Programme; European Research Council; Consejo Superior de Investigaciones Científicas; Novo Nordisk Fonden; National Institutes of Health; European Regional Development Fund; Ministerio de Ciencia e Innovación; European Commission; UK Research and Innovation; National Science Foundation; Grantham Foundation for the Protection of the Environment; Frances and Augustus Newman Foundation; Novo Nordisk; HORIZON EUROPE Framework Programme; Intramural Research Program; Cambridge Trust; Carl-Zeiss-Stiftung","keywords":"Benchmark (surveying); Computer science; Set (abstract data type); Field (mathematics); Computation; Fine-tuning; Work (physics); Simple (philosophy); Range (aeronautics); Machine learning; Artificial intelligence; Engineering; Algorithm; Epistemology; Programming language","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003070191,0.0001801704,0.0002510679,0.0001144302,0.0003773829,0.0005855623,0.00127892,0.00004940313,0.000372816],"category_scores_gemma":[0.0002276606,0.000150092,0.0000419177,0.001363643,0.002657926,0.0007534435,0.0006129974,0.0001093297,0.00001016414],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001728234,"about_ca_system_score_gemma":0.000651073,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00007074963,"about_ca_topic_score_gemma":0.000003940432,"domain_scores_codex":[0.9971088,0.00001706686,0.0003853597,0.00100443,0.0008212658,0.0006630765],"domain_scores_gemma":[0.9987896,0.0001652758,0.0001218088,0.0005045463,0.0001819999,0.0002367333],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.000005730979,0.00003088966,0.00005006099,0.0002271042,7.562522e-7,9.55617e-7,0.0001507365,0.0001095438,0.9837155,0.01522918,0.00003155213,0.0004479367],"study_design_scores_gemma":[0.0001190352,0.00001588895,0.0003069093,0.00006036812,0.000007803261,0.000006618273,0.00004710724,0.08174416,0.9141275,0.003102976,0.0002869314,0.0001747348],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9715837,0.00004400443,0.02114567,0.0001640429,0.0003634186,0.0004358619,0.0001945147,0.0001944113,0.005874413],"genre_scores_gemma":[0.9529306,0.000005724605,0.04647512,0.00002739023,0.0001532744,0.0001579503,0.000009864892,0.00001391229,0.0002261247],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.08163461,"threshold_uncertainty_score":0.979324,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01023883856110066,"score_gpt":0.3383258651224839,"score_spread":0.3280870265613833,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}